Expandable in -vehicle camera application range, high resolution needs

Currently, as a camera mounted on a car, the front sensit, which is responsible for the recognition of the front object while driving, the peripheral sensor that recognizes the surrounding status of the vehicle, including the side and the rear, and monitoring the driver to sleep and the vital. The application range is expanding, such as the indoor car room (in cabin). The resolution of the mainstream cameras of the front sensing and all over -sensing is about 1-2 MP, but what happens if this is a high -resolution? For example, in the case of front sensing, if you want to detect a child riding a 45m -ahead bicycle with a 2MP wide viewing angle camera, it will be about 15 pixels vertically. The conditions change depending on the road surface and weather, but in order to recognize them earlier and realize safe driving, it is necessary to identify further far away, and for that, it is clearly identified at a higher resolution. You need to be able to do it.

For this reason, ONSEMI has been developing that the resolution of 8MP in restricted autonomous driving requires 8MP and 12MP or more at automatic driving level 4 or 5 or 5.Also, in the automotive field, the resolution alone cannot compete.For example, in an era where fully automatic driving cars spread to the market in earnest, it is assumed that a car that is vacant in a smartphone app will be used at any time.In that case, automobiles will always be required to continue to operate, and the extended operation time is estimated to be 8000 hours (in the case of the United States) in one year, so it is necessary to ensure the reliability that can withstand it.Come out.

Seeing that such needs will come out, the company's aiming to improve reliability, for example, the acquisition of deterioration data due to ultraviolet rays of a color filter, such as the search for measures to respond, and the length of the product life cycle.It is said that it is promoting stricter tests that are more strict by car manufacturers for lifespan, function safety as system integration, and security mechanism efforts.

In addition, what kind of error the image sensor causes, and by providing such a problem first to reduce the number of development man -hours and the coverage of defect measures, the cyber security that should have an image sensor should be.It is said that it is also promoting the import of the organization.

Expansion of HDR, which is essential in the car

There are two main ways to improve the resolution of the CMOS image sensor.A method of increasing the number of pixels by increasing the size of the image sensor while keeping the pixel size as it is.The other is to reduce the pixel size and increase the number of pixels while reducing the size of the image sensor as before.It is not realistic to increase the size of the former image sensor because it led to the size of the mounted system and the reduction in the number that can be removed from one wafer.However, in the latter method, the amount of light coming into the pixels decreases, and even if the resolution is increased, there is a problem that the image acquisition conditions in the dark place become stricter, so this is a response in this dark place.Is one of the important points in the in -vehicle image sensor.

Many in -vehicle image sensor makers show them in demonstrations include recognition accuracy in dark underground parking lots and tunnels, and curation control when moving from such a dark place to bright places.AR0820 is 2.1 μm pitch 8.It is realized that 3MP is realized, and high -image quality imaging is possible even in a low -light environment of about 10LUX.But 1.In the darker conditions such as 4Lux, the existing 4.2 μm pitch 1.Compared to the 7MP product, it is still a dark picture.Therefore, the company prepared a pixel addition mode.By combining pixel data between the neighbors, the resolution decreases to 1/4, but 4.The brightness equivalent to 2 μm pitch can be realized.Since the normal mode and pixel addition mode can be used properly for each frame, it is possible to change the mode alternately to combine each strength.

In addition, the aim is to clearly reflect the video, and many of the existing in -vehicle image sensors have a 120dB dynamic range to 140db/24 bits (it can be used by switching between 120db mode and 140dB mode).This makes it possible to clearly reflect the details in a dark environment.

Direction of in -vehicle image sensor in the future

Although it is an in -vehicle image sensor that has a feeling of coming to the point, the development of a next -generation product aims to achieve a higher level that does not already compromise that ONSEMI still has a compromise.It is said that it is proceeding.

As a next -generation product, the pixel size is the same as the AR0820 2.Adopt 1 μm pitch, but 2 new.The company plans to adopt a 1μm Super Exposure pixel.By utilizing this, the current 3 μm and 3.In addition to confirming that the same SNR as the 75 μm pitch process can be achieved, even if you can find a large stone on the street at night, you can grasp more than 150m in a 3 μm pitch of 150m.He also confirmed that it was possible.

In addition, in the case of HDR, when connecting multiple images, the case where SNR deteriorates is the current image sensor, and that HDR is realized with multiple pixels as adopted by competitors., SNR is less than 20dB, but by adopting a device to reduce the decline in HDR transition SNR, it is possible to keep SNR to 25 dB or more even when the junction temperature is 125 ° C.Suppose you can maintain high image quality.

In addition, this next -generation image sensor product has already begun to provide alpha suspension, realizing the support of a dynamic range of up to 155 dB, and providing low illuminance equivalent to the 3 μm pixel image sensor.So, the development will continue as it is.